Re: Nucleus counter: statistics

Posted by Martin du Saire on Oct 21, 2005; 9:58pm
URL: http://imagej.273.s1.nabble.com/Nucleus-counter-tp3704628p3704632.html

Only a few?  Isn't it necessary to determine statistically the
reliability of the automated routine?

Martin

At 09:48 AM 10/21/2005, you wrote:

>Every time I write an automated routine I like to test with computer
>generated data where I know what the answer is and with at least one
>(typically a few, but who has the time?) real biological samples
>where I've done the analysis manually or by another method.
>-Michael
>
>At 12:29 PM 10/19/05 -0500, you wrote:
>>I have a related question:
>>
>>Once one ends up with an algorithm for counting blobs (nuclei,
>>mosquito eggs, etc.)  how should one go about evaluating the
>>efficacy of the procedure, and are there any major statistical
>>pitfalls to avoid?  Just looking for a shove in the right direction.
>>
>>Thanks.
>>
>>Martin
>
>____________________________________________________________________________
>Michael Cammer   Analytical Imaging Facility   Albert Einstein Coll. of Med.
>Jack & Pearl Resnick Campus      1300 Morris Park Ave.     Bronx, NY  10461
>[phone]       Fax:  [phone]      URL:  http://www.aecom.yu.edu/aif/
>   **This electronic transmission contains information that is privileged.**